1. Correlation Chapter 15

2. A research design reminder >Experimental designs You directly manipulated the independent variable. >Quasi-experimental designs You examined naturally occurring groups. >Correlational designs You just observed the independent variable.

3. Defining Correlation >Co-variation or co-relation between two variables >These variables change together >Usually scale (interval or ratio) variables >Correlation does NOT mean causation!

4. Correlation versus Correlation >Correlational designs = research that doesn’t manipulate things because you can’t ethically or don’t want to. You can use all kinds of statistics on these depending on what you types of variables you have. >Correlation the statistic can be used in any design type with two continuous variables.

5. Correlation Coefficient >A statistic that quantifies a relation between two variables >Tells you two pieces of information: Direction Magnitude

6. Correlation Coefficient >Direction Can be either positive or negative Positive: as one variable goes up, the other variable goes up Negative: as one variable goes up, the other variable goes down.

7. Positive Correlation Association between variables such that high scores on one variable tend to have high scores on the other variable A direct relation between the variables

8. Negative Correlation Association between variables such that high scores on one variable tend to have low scores on the other variable An inverse relation between the variables

9. A Perfect Positive Correlation

10. A Perfect Negative Correlation

11. Correlation Coefficient >Magnitude Falls between -1.00 and 1.00 The value of the number (not the sign) indicates the strength of the relation

13. A visual guide

14. Check Your Learning >Which is stronger? A correlation of 0.25 or -0.74?

15. Misleading Correlations >Something to think about There is a 0.91 correlation between ice cream consumption and drowning deaths. >Does eating ice cream cause drowning? >Does grief cause us to eat more ice cream?

16. Another silly example

17. Correlation Overall >Tells us that two variables are related How they are related (positive, negative) Strength of relationship (closer to | 1 | is stronger). >Lots of things can be correlated – must think about what that means.

18. The Limitations of Correlation >Correlation is not causation. Invisible third variables Three Possible Causal Explanations for a Correlation

19. Causation Direction

20. The Limitations of Correlation, cont. >Restricted Range. A sample of boys and girls who performed in the top 2% to 3% on standardized tests - a much smaller range than the full population from which the researchers could have drawn their sample.

21. >Restricted Range, cont. If we only look at the older students between the ages of 22 and 25, the strength of this correlation is now far smaller, just 0.05.

22. The Limitations of Correlation, cont. >The effect of an outlier. One individual who both studies and uses her cell phone more than any other individual in the sample changed the correlation from 0.14, a negative correlation, to 0.39, a much stronger and positive correlation!

23. The Pearson Correlation Coefficient >A statistic that quantifies a linear relation between two scale variables. >Symbolized by the italic letter r when it is a statistic based on sample data. >Symbolized by the italic letter p “rho” when it is a population parameter.

24. >Pearson correlation coefficient r Linear relationship

25. Correlation Hypothesis Testing >Step 1. Identify the population, distribution, and assumptions >Step 2. State the null and research hypotheses. >Step 3. Determine the characteristics of the comparison distribution. >Step 4. Determine the critical values. >Step 5. Calculate the test statistic >Step 6. Make a decision.

27. SPSS Since I have two scores from each person, be sure to line them up side by side.

28. SPSS Under variable view, be sure both are listed as scale – otherwise you won’t be able to create the right type of graph.

29. SPSS Graph > Chart Builder

30. SPSS Click on scatter/dot, and drag/drop the first box into the graph building area.

31. SPSS Drag one variable into X, one variable into Y. Remember, this view is NOT the real scatterplot.

32. SPSS Hit ok!

33. SPSS Starting with a scatterplot helps you to see what the direction (negative) and magnitude (strong) should be for the correlation. Remember that N has a large influence on power for rejecting the null, so even strong correlations can be nonsignificant.

34. SPSS To get the correlation coeffiecient, r: Analyze > correlate > bivariate

35. SPSS Move both variables over to the right, hit ok.

36. SPSS

37. SPSS Output

38. Assumptions >Random selection >X and Y are at least scale variables >X and Y are both normal >Homoscedasticity (for real this time!) Each variable must vary approximately the same at each point of the other variable. See scatterplot for UFOs, megaphones, snakes eating dinner.

39. Hypothesis Steps Step 1: Population: no correlation between (var 1) and (var 2) Sample: correlation between (var 1) and (var 2)

40. Hypothesis Steps Step 2 (with our example): Null: r for exam grades and absences = 0 Research: r for exam grades and absences /= 0 (can also use the rho symbol).

41. Hypothesis Steps Step 3: List r = -.851 List df = N – 2 *** = 8 *** whoa! It’s N – 1(for X) – 1(for Y).

42. Hypothesis Steps Step 4: (note: different from your book) Find the t-cut off score from a t-table given your df. p <.05 = + / - 2.306

43. Hypothesis Steps Step 5: Calculate the t-found given your r value.

44. Or you can use MOTE!

45. Effect Size? >r is often considered an effect size. Most people square it though to r 2, which is the same as ANOVA effect size. You can also switch r to Cohen’s d, but people are moving away from doing this because d normally is reserved for nominal variables.

46. Confidence Intervals >Involves Fisher’s r to z transform and is generally pretty yucky. Which means people do not do them unless they are required to.

47. Correlation and Psychometrics >Psychometrics is used in the development of tests and measures. >Psychometricians use correlation to examine two important aspects of the development of measures—reliability and validity.

48. >A reliable measure is one that is consistent. >Example types of reliability: Test–retest reliability. Split-half reliability >Cronbach’s alpa (aka coefficient alpha) Want to be bigger than.80. Reliability

49. Validity >A valid measure is one that measures what it was designed or intended to measure. >Correlation is used to calculate validity, often by correlating a new measure with existing measures known to assess the variable of interest.

50. >Correlation can also be used to establish the validity of a personality test. >Establishing validity is usually much more difficult than establishing reliability. >Buzzfeed!

51. Partial Correlation >A technique that quantifies the degree of association between two variables after statistically removing the association of a third variable with both of those two variables. >Allows us to quantify the relation between two variables, controlling for the correlation of each of these variables with a third related variable.

52. >We can assess the correlation between number of absences and exam grade, over and above the correlation of percentage of completed homework assignments with these variables.